>Number:         29616
>Category:       pkg
>Synopsis:       db4 still suffers from gcc 3.3 optimization bug on powerpc
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    pkg-manager
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Mar 07 11:14:00 +0000 2005
>Originator:     Timo Schöler
>Release:        NetBSD 2.0/macppc
>Organization:
//macfinity
>Environment:
NetBSD fulcrum.ts39-bln.macfinity.net 2.0 NetBSD 2.0 (GENERIC.MP) #0: Wed Dec  1 14:05:26 UTC 2004  builds@build:/big/builds/ab/netbsd-2-0-RELEASE/macppc/200411300000Z-obj/big/builds/ab/netbsd-2-0-RELEASE/src/sys/arch/macppc/compile/GENERIC.MP macppc
>Description:
as stated in pkgsrc/databases/db4/hacks.mk, this small hack solves PR 23737:

### [Sun Mar 14 19:32:40 UTC 2004 : jlam]
### Pre-3.3 GCC has an optimization bug on powerpc that's tickled by
### the db4 source code, so remove optimization flags in that case.
### This fixes PR 23737 by John Klos.

*it does not*

i built db-4.3.27 yesterday with the on-board compiler of NetBSD 2.0 (which is gcc 3.3.3) and run into a non-functioning cyrus-imapd which stated a cyrus-killing database failure in /var/log/messages.

trying the binary package from ftp.netbsd.org -- it runs. i'm building gcc 3.4.3 right now to see if this solves the problem.

but still, the gcc 3.3.3 issue remains.
>How-To-Repeat:
build db4 with the on-board gcc (which is 3.3.3) and watch it fail (e.g. by using it as backend for the cyrus-imapd).
>Fix:
i) use the binary package;

ii) try using gcc 3.4.3?

iii) wait for the fix